The External Validity of the Sociological Laboratory Experiment

实验室实验作为社会学的一种规范方法还很少被确立，这通常是因为实验室研究结果在现实世界中的可传递性有限。如果受试者样本的概括不仅包括治疗组和对照组之间刺激引起的差异(例如，合作率变化5个百分点)，而且还包括绝对值(例如，50%的合作率)，则违反外部效度的情况尤其严重。实际上，社会学实验的外部效度还没有得到系统的检验。这项资助提案关注的是来自社会学实验室实验的刺激效应和边际总数的可靠性和外部效度。预期结果将允许对以前发现的实验室结果进行确凿的重新评估，并确定相关参数，以改进未来实验室实验研究的设计。利用学生主题库，我们计划在莱比锡和慕尼黑的两个研究地点进行一系列相同的实验室实验(平行测试可靠性)，多次邀请受试者到实验室研究学习和重复效应(重测可靠性)，改变邀请程序的努力以确定参与者自我选择的潜在影响，以及改变参与者之间以及参与者和实验者之间的匿名性(反应性)。此外，我们通过系统地比较实验室和在线结果来测试在线实验的有效性，并通过与具有代表性的德国人口样本(全国范围内的可移植性)和国际参与者进行在线实验来测试实验结果的跨文化概括性，从而大大拓宽了受试库。我们将使用合作研究中的著名决策情景(独裁者博弈、最后通牒博弈、囚徒困境)。选择的游戏涉及社会期望行为的情景，复杂程度各不相同，包含经过良好评估的刺激，并允许与先前的实验室结果进行直接比较。这三种互动情况都具有社会学意义，并为研究公平、制裁和制裁威胁以及合作的准则提供了可能性。

Laboratory experiments are only little established as a canonical method of sociology, which often has been justified by the limited transportability of laboratory findings to the real world. Violations of external validity are particularly serious if generalizations from subject samples not only cover stimulus induced differences between treatment and control groups (e.g. changes in the cooperation rate by 5 percentage points), but absolute values (e.g. a cooperation rate of 50 percent). Effectively, the external validity of sociological experiments has not been tested in a systematic manner. This funding proposal is concerned with the reliability and the external validity of stimulus effects and marginal totals from sociological laboratory experiments. The expected results will allow for a substantiated re-evaluation of previously found laboratory results as well as the identification of relevant parameters to improve the design of future laboratory experimental studies. Employing student subject pools we plan to conduct a series of identical laboratory experiments at two research sites in Leipzig and Munich (parallel test reliability), invite subjects to the laboratory several times in order to study learning and repetition effects (test-retest reliability), vary the effort of the invitation procedure to identify potential effects from self-selection of participants, and vary the anonymity between participants as well as between participants and the experimenter (reactivity). Additionally, we test the validity of online experiments by systematical comparison of laboratory and online results and substantially broaden the subject pool by conducting online experiments with a representative sample of the German population (nationwide transportability) and with international participants to test the cross-cultural generalizability of experimental findings. We will use well-known decision situations from the study of cooperation (dictator game, ultimatum game, prisoners dilemma). The chosen games involve situations of social desirable behavior, vary in complexity, contain stimuli with well assessed effects, and allow for a direct comparison with prior laboratory results. All three situations of interaction are sociologically relevant and offer the possibility to study norms of fairness, sanctions and the threat of sanctions as well as cooperation.