Mill and Landbase Attributes Condition Response Rates to Timber Product Output Surveys in the Western United States

David L.R. Affleck, George C. Gaines

Research output: Contribution to journalArticlepeer-review

Abstract

Since the 1940s, the USDA Forest Service has surveyed roundwood receiving facilities to report on timber products output (TPO). The TPO program provides information on forest removals, processing capacity, and wood product markets, complementing inventory estimates of status and change. The program adopted an annual state-level probability sampling design in 2019 and this research evaluates rates of unit nonresponse (UN) in surveys of eleven western states over reference years 2019–2021. The overall response rate for this region and period was 43.7%, which appears low relative to prior western surveys. Response rates also differed significantly by state, facility type, and facility size class. Furthermore, distributions of forestland ownership and harvest levels over potential facility feedstock areas differed significantly between respondents and nonrespondents for an important class of large sawmills. Taken together, these results suggest UN is substantially degrading the precision of western TPO surveys and may be inducing appreciable bias in current estimators. At the same time, response rates were significantly higher where responses were obtained in previous years, suggesting a positive facilitation effect. The implications of these rates and patterns in UN for TPO survey design and estimation are discussed. Study Implications: The USDA Forest Service estimates timber products output (TPO) from US forests using sample surveys of roundwood receiving facilities. In TPO samples of western states for the period 2019–2021, the average response rate was only 43.7%. The level of unit nonresponse will substantially degrade the precision of TPO estimates at state and regional levels, while the observed patterns may induce bias. Further research is needed to identify the bias and precision of current and alternative estimators and to investigate survey implementation factors or design enhancements that would allow more intensive follow-up efforts to focus on a smaller number of nonrespondents.

Original languageEnglish
Pages (from-to)14-22
Number of pages9
JournalForest Science
Volume70
Issue number1
DOIs
StatePublished - Feb 1 2024

Keywords

  • establishment surveys
  • nonresponse bias
  • probability sampling
  • unit nonresponse

Fingerprint

Dive into the research topics of 'Mill and Landbase Attributes Condition Response Rates to Timber Product Output Surveys in the Western United States'. Together they form a unique fingerprint.

Cite this