Assessing the point at which averages are stable: A tool illustrated in the context of person perception

Across many diverse areas of research, it is common to average a series of observations, and to use these averages in subsequent analyses. Research using this approach faces the challenge of knowing when these averages are stable. Meaning, to what extent do these averages change when additional observations are included? Using averages that are not stable introduces a great deal of error into any analysis. The current research develops a tool, implemented in R, to assess when averages are stable. Using a sequential sampling approach, it determines how many observations are needed before additional observations would no longer meaningfully change an average. The utility of this tool is illustrated in the context of impression formation, demonstrating that averages of some perceived traits (e.g., happy) stabilize with fewer observations than others (e.g., assertive). A tutorial regarding how to utilize this tool in researchers’ own data is provided.

在众多研究领域中，对一系列观测值进行平均并利用这些平均值进行后续分析是一种常见的做法。采用此方法的研究面临着一个挑战，即确定这些平均值何时稳定。换言之，在包含额外观测值时，这些平均值的变化程度如何？使用不稳定的平均值将导致分析中引入大量的误差。当前的研究开发了一种工具，该工具采用 R 语言实现，用于评估平均值何时稳定。通过采用顺序抽样方法，它确定了在额外观测值不再有意义地改变平均值之前，需要多少观测值。该工具的实用性通过印象形成的语境进行说明，展示了某些感知特征的平均值（例如，快乐）在比其他特征（例如，自信）需要更少的观测值时即可稳定。同时，提供了一份教程，指导研究人员如何在自己的数据中运用此工具。