Many studies in master psychology and professional psychology study programs used quasi-experimental methods, but there was no reference regarding quasi-experiments written in Indonesian. This article will fill in the blanks on reference to quasi-experimental methods. The article explains quasi-experimental design or non-randomized experimental design. According to Campbell quasi experimental design is divided into four types, namely (a) quasi-experimental design without control group or pretest, b) quasi experimental design with control group and pretest, c) time series design, and d) regression discontinuity design. Each type was broken down into a more specific design.

In Experiment 1, 4 rats earned their daily food ration by choosing on a trials basis between a "risky" and a "riskless" lever. The risky lever produced either 15 45-mg food pellets or no pellets, and on average provided five pellets per choice. The riskless lever always produced three pellets. Across conditions, the number of trials per session was varied. Body weight and choice of the risky lever decreased as the number of trials per session decreased, even though body weight could only be defended by increased choice of the risky lever. In Experiment 2, trials per session were fixed, but the number of pellets delivered by the risky and riskless levers was either at the same level as in Experiment 1 or tripled from those levels. Now choice of the risky lever was inversely related to the size of reinforcement and to body weight. The results of these experiments show that risk aversion covaries with the amount of food available in a session and the daily variance in the amount of food earned.

Three rats earned their daily food ration by responding during individual trials either on a lever that delivered one food pellet immediately or on a second lever that delivered three pellets after a delay that was continuously adjusted to ensure substantial responding to both alternatives. Choice of the delayed reinforcer increased when the number of trials per session was reduced. This result suggests that models seeking closure on choice effects must include a parameter reflecting how preference changes with sessionwide income. Moreover, models positing that reinforcer probability and immediacy (1/delay) function equivalently in choice are called into question by the finding that probability and immediacy produce opposing effects when income level is changed.

In Experiment 1, 4 rats earned their daily food ration by choosing between two levers. One lever delivered two regular and one quinine-adulterated food pellets, and the other delivered two regular and four quinine pellets. A 20-s intertrial interval separated successive choices. Sessions began with 10 forced trials during which only one lever, selected with p = .5 and cued by a light above it, could deliver its reinforcer. Forced trials were followed by 30 or 150 trials, depending on the condition, during which choices to either lever could be reinforced. Over this range, absolute choice of the fourquinine, two-regular-pellet lever was inversely related to the number of free-choice trials, establishing this reinforcer as an inferior good. In Condition 1 of Experiment 2, the prior design was altered in two ways: (a) one lever delivered four quinine pellets, and the other lever delivered one standard pellet; and (b) sessions ended after 140 free-choice trials. When the number of free-choice trials was reduced to 100 (Condition 2), all 3 rats increased their preference for quinine pellets, confirming their status as an inferior good. In the next several conditions, the number of quinine pellets provided for selecting its associated lever was varied between three and four. Preference for the quinine-pellet alternative was inversely related to the number of pellets it provided, a result defining it as a Giffen good. These findings are not accommodated readily by extant choice models and complicate the search for a unitary model of choice.

Pengantar 1Rancangan eksperimen oleh McGuigan (1987) secara umum diartikan sebagai penerapan metode ilmiah yang diawali dengan merumuskan permasalahan, dilanjutkan dengan merumuskan hipotesis, menyeleksi partisipan, menempatkan partisipan ke kelompok eksperimen dan kontrol, menentukan variabel independen dan dependen, mengendalikan variabel luar yang relevan, melakukan analisa statistik, membuat generalisasi dan penjelasan hipotesis jika terkonfirmasi, serta diakhiri dengan memprediksi terhadap situasi baru lewat replikasi. Rumusan rancangan eksperimen versi McGuigan ini merupakan rumusan yang luas sebab halhal diatas merupakan penerapan tahapan seseorang didalam menyusun penelitian. Pengertian rancangan eksperimen secara lebih sempit dikemukakan oleh Kirk (1982)