PARKER, Justice.
We granted certiorari in this workers’ compensation case to review an alleged conflict between the no-opinion affirmance of the trial court’s judgment by the Court of Civil Appeals and that court’s earlier decisions regarding the appropriate application of § 25-5-57(b), Ala.Code 1975, a portion of the Workers’ Compensation Act, § 25-5-1 et seq., Ala Code 1975. Section 25-5-57(b) provides:
“(b) Computation of compensation; determination of average weekly earnings. Compensation under this section shall be computed on the basis of the average weekly earnings. Average weekly earnings shall be based on the wages, as defined in Section 25-5-1(6)[1] of the injured employee in the employment in which he or she was working at the time of the injury during the period of 52 weeks immediately preceding the date of the injury divided by 52, but if the injured employee lost more than seven consecutive calendar days during the period, although not in the same week, then the earnings for the remainder of the period, although not in the same week, then the earnings for the remainder of the 52 weeks shall be divided by the number of weeks remaining after the time so lost has been deducted. Where the employment prior to the injury extended over a period of less than 52 weeks, the method of dividing the earnings during that period by the number of weeks and parts thereof during which the employee earned wages shall be followed, provided results just and fair to both parties will thereby be obtained. Where by reason of the shortness of the time during which the employee has been in the employment of his or her employer or the casual nature or terms of the employment it is impracticable to compute the average weekly earnings as above defined, regard shall be had to the average weekly amount which during [891]*891the 52 weeks prior to the injury was being earned by a person in the same grade, employed at the same work by the same employer, and if there is no person so employed, by a person in the same grade employed in the same class of employment in the same district. Whatever allowances of any character made to an employee in lieu of wages are specified as part of the wage contract shall be deemed a part of his or her earnings.”
We conclude that a conflict does exist, and we reverse the judgment of the Court of Civil Appeals and remand the case to that court for resolution.
Facts
Barbara Ann Gibson began work for Dollar General, a retail discount store owned by Dolgencorp, Inc. (“Dolgen”), as an hourly wage employee on or about December 2,1998. She continued to work for Dolgen for 38 weeks2 until she suffered a work-related injury on July 28, 1999. She had become store manager in June 1999, and her salary was increased to $425 per week. After her injury, Gibson resigned. Her earnings for the 33 weeks that she worked totaled $8,715.88. This amount divided by 33 weeks results in average weekly earnings of $264.12.
Gibson sued Dolgen in the Clarke Circuit Court in a dispute over the extent of her disability and the amount of her average weekly earnings, which would determine the amount of any disability payments. The extent of Gibson’s disability is not at issue on appeal, but Dolgen challenges the trial court’s finding that Gibson’s average weekly earnings were $425. The trial court, apparently relying heavily on the fact that Dolgen reported Gibson’s salary as $425 per week when it filed the “Employer’s First Report of Injury,” used that amount as the basis for computing Gibson’s benefits.
The trial court’s order, dated December 13, 2005, reads, in pertinent part, as follows:
“This cause came before the Court on the 27th day of June, 2005, upon the pleadings, the evidence ore tenus adduced at trial,[3] and the arguments of counsel. Upon consideration of the same, the Court finds as follows:
[[Image here]]
“The parties are in dispute regarding two material issues:
“1. The calculation of [Gibson’s] average weekly earnings. Dolgencorp, Inc. contends that the second method of [Ala.] Code [1975,] § 25-5-57(b) should be used to calculate [Gibson’s] average weekly earnings, which would result in the amount of $264.12. [Gibson] contends that the third method of [Ala.] Code [1975,] § 25-5-57(b) should be used to calculate her average weekly earnings, which would result in the amount of $425.00. Four Hundred and Twenty-five dollars ($425.00) is also the average weekly wage self-reported by Dolgencorp, Inc. to the State of Alabama [892]*892in Employer’s First Report of Injury. This issue was briefed extensively by the parties, evidence was submitted on the issue, and counsel for both parties argued the issue.
[[Image here]]
“Findings and Conclusions
“Upon consideration of the medical evidence, the vocational evidence, and the other evidence ore tenus, including the Court’s observations of Barbara Ann Gibson, the Court has arrived at the following findings of fact and conclusions of law:
“1. That calculating Barbara Ann Gibson’s average weekly earnings using the second method of [Ala.] Code [1975,] § 25-5-57(b) would not obtain a just and fair result for both parties.
“2. That it is impracticable to compute the average weekly earnings in this case using the second method of [Ala.] Code [1975,] § 25-5-57(b).
“3. That calculating [Gibson’s] average weekly earnings using the third method of [Ala.] Code [1975,] § 25-5-57(b) is practicable in obtaining a result that is just and fair to both parties.
“4. That the overwhelming evidence in support of the calculation of [Gibson’s] average weekly earnings using the third method of [Ala.] Code [1975,] § 25-5-57(b) obtains a result of $425.00 per week.
“5. That the result of $425.00 per week calculating [Gibson’s] average weekly earnings using the third method of [Ala.] Code [1975,] § 25-5-57(b) is equal to the average weekly wage self-reported by Dolgencorp, Inc. to the State of Alabama in Employer’s First Report of Injury.”
Dolgen appealed to the Court of Civil Appeals, challenging only the trial court’s finding of average weekly earnings of $425, arguing that the trial court should have calculated Gibson’s benefit based on average weekly earnings of $264.12 under the statute, and not on average weekly earnings of $425. The Court of Civil Appeals affirmed the trial court’s judgment, without an opinion, on October 20, 2006, citing only Henderson v. Johnson, 632 So.2d 488, 490 (Ala.Civ.App.1993), presumably that portion of Henderson that states that “when it is impracticable to apply the formulas for determining average weekly wage so as to arrive at a just and fair result to both parties, it is left to the sound judgment and judicial discretion of the trial court.” Dolgencorp, Inc. v. Gibson (No. 2050335, Oct. 20, 2006), 3 So.3d 296 (Ala.Civ.App.2006) (table). Dolgen sought cer-tiorari review of the Court of Civil Appeals’ decision, arguing that it conflicts with several decisions of this Court and of the Court of Civil Appeals. This Court granted Dolgen’s petition.
Dolgen argues that in Collins v. Westmoreland, 600 So.2d 253, 255 (Ala.Civ.App.1991), the Court of Civil Appeals construed § 25-5-57(b), Ala.Code 1975, to mean that where the duration of employment is less than 52 weeks, the second method in the statute for determining average weekly earnings should be used. Dolgen argues that the Court of Civil Appeals’ decision to affirm the judgment of the trial court here conflicts with that earlier opinion. Gibson argues that the trial court was within its discretion to determine her average weekly earnings as it did, because, Gibson asserts, the statute is not mandatory when the employee, as here, did not work in the employment for 52 weeks preceding the injury. Gibson also argues that Dolgen’s insertion of $425 as the average weekly salary on the Employer’s First Report of Injury is an admission against interest that supports the trial court’s finding even [893]*893where other evidence might support a contrary finding. Gibson’s brief at 27.
Standard of review
“On certiorari review, this Court accords no presumption of correctness to the legal conclusions of the intermediate appellate court. Therefore, we must apply de novo the standard of review that was applicable in the Court of Civil Appeals.” Ex parte Toyota Motor Corp., 684 So.2d 132, 135 (Ala.1996). The standard of review in the Court of Civil Appeals was as follows:
“[An appellate court] will not reverse the trial court’s finding of fact if that finding is supported by substantial evidence — if that finding is supported by ‘evidence of such weight and quality that fair-minded persons in the exercise of impartial judgment can reasonably infer the existence of the fact sought to be proved.’ ”
Ex parte Trinity Indus., Inc., 680 So.2d 262, 268-69 (Ala.1996) (quoting West v. Founders Life Assurance Co. of Florida, 547 So.2d 870, 871 (Ala.1989)). However, “an appellate court’s review of the proof and consideration of other legal issues in a workers’ compensation case shall be without a presumption of correctness.” Ex parte American Color Graphics, Inc., 838 So.2d 385, 387-88 (Ala.2002) (citing § 25-5 — 81(e)(1), Ala.Code 1975). Ex parte Southern Energy Homes, Inc., 873 So.2d 1116, 1121 (Ala.2003).
As previously noted, no ore tenus evidence was presented to the trial court regarding this issue. Our review, therefore, is de novo, requiring no deference to the trial court as would be the case under the ore tenus rule.
Analysis
“The Legislature enacted the schedule in the Workers’ Compensation Act with a primary purpose of minimizing costly and time-consuming litigation over work-related injuries.” Ex parte Addison Fabricators, Inc., 989 So.2d 498, 502-03 (Ala.2007).
“The Workers’ Compensation Law was adopted in deference to a widespread belief and demand that compensation should be awarded to workers injured and disabled temporarily or permanently in the course of the employment, even though sometimes the accident might occur under such circumstances as would not permit a recovery in an ordinary action at law. The underlying thought was that such a system of compensation would be in the interest of the general welfare by preventing a worker from being deprived of means of support as the result of an injury received in the course of employment. The statute is the expression of what was regarded by the Legislature as a wise public policy concerning injured employees.”
1 Terry A. Moore, Alabama Workers’ Compensation § 2:3 (1998) (footnote omitted).
“[The Workers’ Compensation Act] has as its beneficial and humanitarian purpose to afford to injured workers and their dependents a means of maintenance during periods of disability where that disability arose from the hazards of the employment. A primary purpose of workers’ compensation is to financially aid the employee and his or her dependents for earnings lost by the employee’s injury. The [Alabama Workers’] Compensation Act has thus been construed to have as its one overriding purpose the protection of injured workers from the devastating effect of personal injuries arising out of and in the course of employment.”
[894]*894Id. at § 2:5 (footnotes omitted)(emphasis added).
As expressed above, one purpose of the Workers’ Compensation Act is to provide the injured worker a means of maintenance during periods of disability, by providing compensation based on his or her average weekly earnings for the prior 52 weeks, or for the number of weeks the employee worked in his or her employment, if less than 52 weeks. The methodology used in providing this maintenance level of compensation is a major consideration in this case.
Dolgen argues that § 25-5-57(b) requires that where the duration of employment was less than 52 weeks, the second method in the statute for determining average weekly earnings should be used. That portion of the statute reads as follows:
“Where the employment prior to the injury extended over a period of less than 52 weeks, the method of dividing the earnings during that period by the number of weeks and parts thereof during which the employee earned wages shall be followed, provided results just and fair to both parties will thereby be obtained.”
§ 25-5-57(b), Ala.Code 1975. Dolgen argues that the Court of Civil Appeals’ decision conflicts with its earlier opinion, Collins, supra, which, Dolgen says, clarified the statutory requirement and requires the trial court to use the second method unless it determines that that method will not produce a fair and just result. Collins states, in pertinent part:
“It has been held that the second method of dividing the employee’s earnings by the number of weeks and parts of weeks employed is not mandatory but that it allows the trial court to determine if its use will provide just and fair results. However, this is not an optional method and should be used unless the trial court determines a valid reason that its use will not produce fair and just results.”
Collins, 600 So.2d at 255 (emphasis added). “Unless the court determines that for some valid reason the use of such method [dividing the earnings by the number of weeks] will not produce just and fair results, the method is not merely optional and should be used.” Odell v. Myers, 52 Ala.App. 558, 563, 295 So.2d 413, 416 (1974) (citing Brunson Milling Co. v. Grimes, 267 Ala. 395, 103 So.2d 315 (1958)). Dolgen argues that “the trial court failed to make any factual finding of any valid reason why the use of the statutory method would produce substantial unfairness or unjust results to the parties.” Petition at 6 (emphasis in original). See § 25-5-88, Ala.Code 1975 (noting that “judgment shall be entered ... and shall contain a statement of the law and facts and conclusions as determined by said judge”). In Brunson Milling, a case analogous to this one, this Court determined:
“Appellee was employed by appellant for only 24 weekly pay periods prior to his accident and is therefore not within such mandatory provision, but this case is brought within that portion of the same subdivision which reads:
“ ‘Where the employment prior to the injury extended over a period of less than fifty-two weeks, the method of dividing the earnings during that period by the number of weeks and parts thereof which the employee earned wages shall be followed, provided results just and fair to both parties will thereby be obtained.’
“The method of computing compensation provided in this subdivision is not mandatory, but gives the trial court the right to determine if said method will produce just and fair results to both [895]*895parties. Garrison v. Woodward Iron Co., 210 Ala. 45, 97 So. 64, 65 [ (1923) ]; H.C. Price Co. v. Lee, [249 Ala. 230, 30 So.2d 579 (1947) ]; Deaton Truck Line v. Acker, 261 Ala. 468, 476, 74 So.2d 717 [(1954)].
“While some discretion is allowed the court under this subsection, the method therein established is not merely optional, but should be used unless substantial unfairness or injustice will result. We find no indication of an intention of the legislature to extend to the trial court the power whereby the court, after computing the compensation as prescribed in said section, can then adjust for a raise in pay received by the injured employee on the last week of his employment next preceding his injury, as was done by the court in the case before us. We conclude that the court exceeded its discretionary power in the way compensation was computed. We think the computation should be made in accordance with the principles herein-above stated.”
267 Ala. at 399, 103 So.2d at 317 (emphasis added). Here, the trial court determined to use the third method to determine Gibson’s average weekly earnings, but, as shown by the “findings and conclusions” section of its order, it presents no facts or reasoning in support of that determination. Both the statute and the applicable case-law require a finding by the trial court that a “just and fair” result requires deviation from the second method, which “is not an optional method and should be used unless the trial court determines a valid reason that its use will not produce fair and just results.” Collins, 600 So.2d at 255.
Gibson argues that because she did not work for Dolgen for 52 weeks preceding her injury the use of the statutory formula is not mandatory. In her brief in the trial court and again in her brief to this Court, Gibson argues that the first method in § 25-5-57(b) for calculating the average weekly earnings applies only to employees who were in their employment for 52 weeks before the injury and is thus inapplicable to her case. Gibson’s brief at 12. We agree.
Gibson then argues that the use of the second method would not obtain a just and fair result for both parties, that it would be impracticable to calculate her earnings using the second method, and that calculating the earnings using the third method is practicable in obtaining a result that is just and fair to the parties. Gibson’s brief at 13. As authority, Gibson cites only the trial court’s decision that forms the basis for our review. The record, the trial court’s order, and Gibson’s brief, however, present no caselaw or reasoning in support of the claim that the use of the second method is impracticable, or of the claim that a calculation using the second method would not obtain a result that is just and fair to both parties. Gibson merely quotes this Court’s statement that the Workers’ Compensation Act, being remedial in nature, should be liberally construed in favor of the employee when reasonable doubts exist. See Ex parte Byrom, 895 So.2d 942, 946 (Ala.2004); Riley v. Perkins, 282 Ala. 629, 213 So.2d 796 (1968). Gibson’s brief at 13. She has presented no argument or evidence, however, that such reasonable doubts exist.
It is true that,
“in determining legislative intent from the language used in a statute, we may not add or detract from that language and ‘[w]hen [that] language is clear, there is no room for judicial construction.’ Water Works & Sewer Bd. of Selma, 833 So.2d at 607. See also, e.g., Alabama Indus. Bank v. State ex rel. Avinger, 286 Ala. 59, 62, 237 So.2d 108, [896]*896111 (1970) (“When [statutory] language is plain, it should be considered to mean exactly what it says.’).”
Limestone County Water & Sewer Auth. v. City of Athens, 896 So.2d 531, 537 (Ala.Civ.App.2004). Section 25-5-57(b) provides for a liberal construction of its terms by requiring the use of the second method of calculation only where a result just and fair to both parties will thereby be obtained. This statutory provision leaves it for the “ ‘sound judgment and judicial discretion of the trial court’ ” to determine those instances where the statutory formula cannot be applied to determine the average weekly wage. See Ex parte Fryfogle, 742 So.2d 1258, 1261 (Ala.1999) (quoting Aluminum Workers Int’l v. Champion, 45 Ala.App. 570, 574, 233 So.2d 511, 514 (1970)).
The proper exercise of judicial discretion, however, is “ ‘ “the exercise of judicial judgment, based on facts and guided by law, or the equitable decision of what is just and proper under the circumstances.” ’ Shiv-Ram, Inc. v. McCaleb, 892 So.2d 299, 321-22 (Ala.2003) (Houston, J., dissenting and quoting Black’s Law Dictionary 467 (6th ed.1990)).” Ex parte Wilson, 984 So.2d 1161, 1175 (Ala.2007) (Cobb, C.J., dissenting). The statute itself provides guidance regarding the method to be used in the event the less-than-52week employment period is too short or too casual in nature to obtain a just and fair result using the second method. The third method the trial court elected to use here provides for the use of the “average weekly amount which during the 52 weeks prior to the injury was being earned by a person in the same grade, employed in the same work by the same employer .... ” § 25-5-57(b).
The term “too short” is not defined in the statute, but the Court of Civil Appeals has analyzed several sets of circumstances in that regard and has upheld the use of the third method in several cases. Dolgen discusses these decisions as follows:
“When the statute speaks of employment being so short or casual in nature that it is impracticable to use the statutorily preferred method, it really means short and casual such that an alternate means of determining the average weekly wage must be made, and the cases where an alternate method has been used and approved demonstrate why. See, for example: Henderson v. Johnson, 632 So.2d 488 (Ala.Civ.App.1993) (worker employed just 7 weeks before injury); Shields v. GTI Corp., 607 So.2d 253 (Ala.Civ.App.1992), cert. denied, 622 So.2d 939 (Ala.1993) ([worker] employed only 2 days before injury); Stevison v. Qualified Personnel, Inc., 571 So.2d 1178 (Ala.Civ.App.1990) ([worker] employed less than one week before injury); C.E. Adams & Co. v. Harrell, 257 Ala. 25, 57 So.2d 83 (1952) ([worker] employed ‘about 3 days’ before injury).”
Dolgen’s brief at 11-12.
Dolgen directs our attention to Meinhardt v. SAAD’S Healthcare Servs., Inc., 952 So.2d 368 (Ala.Civ.App.2006), a case in which the Court of Civil Appeals discussed what length of employment is sufficient to justify using the second method. In Mein-hardt, the employee, Meinhardt, had been employed for 35 weeks before the injury. The Court of Civil Appeals decided as follows:
“The calculation method ... dividing the earnings by the number of weeks of employment, is appropriate given the length of Meinhardt’s employment. Although § 25-5-57(b) goes on to address circumstances in which the ‘shortness of time’ during which the employee has been employed or ‘the casual nature or terms of the employment’ makes it ‘impracticable’ to compute the average [897]*897weekly earnings, the • circumstances of this case do not support a deviation from the [second] calculation method .... Meinhardt’s 35-week period of employment was not so short, nor so casual in nature, as to make it ‘impracticable’ to compute her average weekly earnings in accordance with the [second method]. The record reveals that Meinhardt chose to work fewer hours on average than other employees during the 35-week period. It is both ‘just and fair’ to calculate Meinhardt’s average weekly wage based on the amount of time she worked. Therefore, we reverse the judgment of the trial court on this issue.”
952 So.2d at 378.
Gibson argues that Meinhardt did not alter the rule affording the trial court discretion to calculate the average weekly earnings using the third method “when it finds, based on substantial evidence, that use of the second method would be unjust, unfair, or impracticable.” Gibson’s brief at 21-22. Gibson overlooks the statutory preference for the second method expressed in the requirement that deviation from that method requires a finding that the result produced using that method would not be just and fair to both parties. No such finding was explicitly stated in the trial court’s decision.
Gibson goes on to argue that Meinhardt is clearly distinguishable because, unlike Gibson, Meinhardt had not experienced a major increase in earnings, Meinhardt had chosen to work fewer hours on average than other employees during the 35-week period, and the calculation of Meinhardt’s average earnings using the second statutory method would be just and fair. Gibson does not elaborate on the alleged significance that should be attached to her promotion to a steady weekly salary, when it “is clear from the plain language of the statute that the average weekly wage is to be determined retrospectively.” W.W. Dyar & Sons, Inc. v. Cochran, 693 So.2d 527, 529 (Ala.Civ.App.1997). Further, it would appear that Gibson, too, worked irregular hours, working less than 40 hours in 16 of the weeks she worked prior to her promotion. There is no evidence presented as to whether the shorter hours were her choice or Dolgen’s choice, but Gibson did not argue before the trial court that a portion of her employment was casual. Gibson worked for 33 weeks before her injury and Meinhardt worked 35 weeks. The periods are comparable, and they contrast sharply with the cases cited above that justified deviation from the second method in cases in which the duration of employment ranged from two days to seven weeks.
Contrary to Gibson’s stated position that the statutory methods are not mandatory, the statute does use the mandatory word “shall” when it provides for the method of calculating average weekly wages for employees who were not in their employment for 52 weeks before their injury. It says: “Where the employment ... extended over ... less than 52 weeks, the method of dividing the earnings during that period by the number of weeks ... during which the employee earned wages shall be followed ....” § 25 — 5—57(b), Ala. Code 1975 (emphasis added). Gibson apparently posits that the wording that follows — “provided results just and fair to both parties will thereby be obtained”— renders the second method optional because of her unsupported assertion that in her case the method would not obtain just and fair results for both parties. We disagree. Short of a demonstration that the results would not be just and fair to both parties, the second method is to be used. “Unless the court determines that for some valid reason the use of such method [898]*898will not produce just and fair results, the method is not merely optional and should be used.” Odell, 52 Ala.App. at 563, 295 So.2d at 416. We hold that if the trial court finds that such a reason exists, it should be presented as justification for the departure from the second method for calculating average weekly earnings for employees with less than 52 weeks’ tenure. Here the trial court presented no reason or reasoning, and the Court of Civil Appeals erred in deferring to the trial court where no deference was warranted under the applicable standard of review.
Our search of the record reveals no finding of fact on the part of the trial court that would require deference on the part of an appellate court, nor do we find in the trial court’s order any evidence indicating that the court found a valid reason that the use of the second method to determine Gibson’s average weekly earnings would not produce a just and fair result. Because such a reason is required by Collins and by the statute, we hold that the Court of Civil Appeals erred in affirming the trial court’s order, which did not meet the requirements that the Court of Civil Appeals presented in Collins.
Gibson raises several other issues that bear consideration, and because we will not reverse the trial court’s judgment if that judgment is based on a finding of fact that is supported by substantial evidence, we have reviewed all aspects of the record in search of such evidence.
Gibson appears to argue that, if the second method is used to calculate her average weekly earnings, the court should consider only her period of employment from the time of her promotion to the time of her injury, stating that her
“lower irregular wages prior to her promotion cannot reasonably be said to provide a ‘just and fair’ basis for calculation of her average weekly earnings. Under such circumstances, therefore, calculation of the employee’s average weekly earnings pursuant to the second method would have been impracticable, unjust, and unfair, and the trial court was well within its discretion in so finding.”
Gibson’s brief at 11. Not only does this argument confuse the standards of the second and third methods of calculating average weekly earnings, but Gibson has provided no factual or legal foundation to support her argument that the wages earned for 24 of the 33 weeks she was employed by Dolgen should not be considered in calculating her average weekly earnings under the statute. Furthermore, the method Gibson insists should be used requires that the “shortness of the time during which the employee has been in the employ of his or her employer ” must make it impracticable to use the other methods, not the shortness of time in a particular position.
We note that the statutory scheme may result in differences between a new hire to a position versus an employee promoted to that position. The third method of calculating average weekly earnings under the statute provides that the average weekly earnings of a newly hired manager injured after a few days on the job would be calculated by using the wages of a surrogate who had worked in the same grade and performed the same work for the same employer for 52 weeks, while the average weekly earnings of a similarly injured newly promoted manager would be calculated by including her earnings in the her lesser-paid position under the second method. Such dissimilarities were the subject of oral argument, but we find no reasoned resolution of the apparent disparity within the confines of the statute. The legislature alone has the authority to correct the statute to alleviate the possibility of any such divergent results.
[899]*899In her brief, Gibson characterizes the issue as whether the trial court exceeded its discretion in calculating the average weekly earnings as it did. This is a question that the Court of Civil Appeals may have considered. We note, however, that “[a] judge abuses his discretion only when his decision is based on an erroneous conclusion of law or where the record contains no evidence on which he rationally could have based his decision.” Dowdy v. Gilbert Eng’g Co., 372 So.2d 11, 12 (Ala.1979) (citing Premium Serv. Corp. v. Sperry & Hutchinson, Co., 511 F.2d 225 (9th Cir.1975)). Based on the guidance provided in Collins, on the Meinhardt decision, on the wording of the statute, and on the dearth of evidence presented at trial, it appears that the trial court exceeded its discretion.
Gibson argues that the fact that she was making $425 per week at the time she was injured, as documented by the Employer’s First Report of Injury filed by Dolgen, provides justification for using the third method to calculate her average weekly earnings. Gibson’s brief at 24. She uses Dolgen’s entry in the Employer’s First Report of Injury as evidence of an average weekly salary of $425. That entry was made in a section of the form that asked if the injured employee had returned to work (yes), if so, the date (8/24/1999), at what wage ($425), at what occupation (store manager), length of time in the employment (9 months), length of time in present job (9 months), average weekly wage ($425), and whether the employee received full pay for day of injury (yes). No evidence was presented to indicate the period for which the “average weekly wage” was sought on the form. Gibson’s average weekly wage at the time of her injury was $425, and that amount had been her average weekly wage since she was promoted. It is incorrect to say that $425 was the average weekly wage for the period of her total employment when the overwhelming evidence presented to the trial court in the form of payroll records shows otherwise. Gibson argues that the entry on the Employer’s First Report of Injury should justify the decision of the trial court under the authority of Garden City Heading Co. v. Thomas, 225 Ala. 273, 142 So. 534 (1932), a case that has been cited one time — in 1955. In Garden City Heading Co., this Court affirmed the judgment of the trial court, stating:
“[Wjhile the time sheets and earnings as set out upon the trial may have shown that the plaintiff had not earned as much as $15 per week for the year preceding the injury, yet the manager of the defendant, Kinser, reported to the ‘Workmen’s Compensation Commission’ that the average weekly earnings of the plaintiff for the past 52 weeks was $15 a week, and this was a fact or inference to support the finding of the trial court.”
225 Ala. at 273, 142 So. at 534. In Garden City Heading Co. the manager testified to the $15 salary as being paid for the preceding 52 weeks, and the trial court accepted his testimony over the written records. However, at the time Garden City Heading Co. was decided, the statute did not define the term “average weekly earnings” as the current statute does. See § 7550(g), Ala.Code 1923. The current statute defines average weekly earnings by reference to § 25-5-1(6), which provides that “average weekly earnings [shall be] based on those earnings subject to federal income taxation and reportable on the Federal W-2 tax form .... ” This would mean a retrospective focus on actual earnings, rather than an annualized projection of most recent earnings. Because the statute requires the use of the income reportable on the federal W-2 tax form, use of the information provided on the employer’s First Report of Injury is not an option.
[900]*900
Conclusion
The Court of Civil Appeals affirmed the trial court’s judgment on the basis that the determination of Gibson’s average weekly earnings should be left to the trial court’s sound judgment and judicial discretion. However, the exercise of judicial discretion is based on facts and guided by law or what is just and proper under the circumstances. In 1869, Justice Peters of this Court discussed judicial discretion, saying:
“In further illustration of what judicial discretion ought to be, but not unfre-quently is not, I add an extract from an opinion, in a case of national importance, by one of our country’s greatest men, and ablest and purest judges. Chief Justice Marshall says: ‘Courts are mere instruments of the law, and can will nothing. When they are said to exercise a discretion, it is a mere legal discretion; a discretion to be exercised in discerning the course prescribed by law; and when that is discerned, it is the duty of the court to follow it. Judicial power is never exercised for the purpose of giving effect to the will of the judge; always for the purpose of giving effect to the will of the legislature; or, in other words, to the will of the law.’ — Osborn v. United States Bank, [22 U.S.] 9 Wheat. 738, [766] [ (1824) ].”
Ex parte Chase, 43 Ala. 303, 311 (1869). Here, the statute provides the course prescribed by law, giving effect to the will of the legislature. It requires that in circumstances such as those presented in this case the second method of calculating average weekly earnings should be used, provided that the results will be just and fair to both parties. Deviation from the second method in favor of the third method requires the documentation of a reasoned and valid finding that the second method would not provide a result that is just and fair to both parties. No such finding was provided in this case; consequently, the decision of the Court of Civil Appeals must be reversed and the matter remanded for reconsideration in accordance with this opinion.
REVERSED AND REMANDED.
SEE, LYONS, WOODALL, STUART, SMITH, and BOLIN, JJ., concur in the result.
COBB, C.J., dissents.
MURDOCK, J., recuses himself.