Methodology

Construction cost estimates were generated by RemodelMAX, a publisher of estimating tools for remodelers, using Clear Estimates remodeling software. The company prepares a detailed construction estimate for each project, and then adjusts this baseline cost for each city to account for regional pricing variations. Construction cost figures include labor, material, and subtrade expenses, plus industry-standard overhead and profit. However, project costs are based on estimates for generic projects and do not account for personalized design and product choices made by homeowners in connection with actual remodeling projects, nor do they account for local and short-term fluctuations in pricing, supply and demand, and other factors affecting cost.

Resale value data for each project was aggregated from estimates provided by real estate professionals via an online survey containing project descriptions and three-dimensional illustrations, plus construction costs and median home prices for each city. Respondents were instructed not to make judgments about the motivation of the homeowner in either the decision to undertake the remodeling project or to sell the house. The survey for this 2016 data was in the field for approximately 10 weeks between August and October, 2015. Nearly 4,000 real estate pros participated.

Margin of Error

The value portion of the Cost vs. Value report is based on a survey. For the national numbers, our statistical partner at The Farnsworth Group has set the confidence interval at 95% +/- 2. This means that if the survey were repeated, 95 times out of 100 the results for any project’s cost-value ratio would be within 2 percentage points of the numbers we published. For instance, our 91.5% ratio for a midrange garage door could go as low as 89.5% or as high as 93.5%. Because the sample sizes for each of the nine census regions are smaller, the potential margin of error is wider. Confidence intervals for each of the nine regions are shown below. The sample sizes of individual markets are too small for us to give margin of error confidence ratings.