Understanding Measurement Error in Agriculture

Last registered on June 15, 2020

View Trial History

Pre-Trial

Trial Information

General Information

Title

Understanding Measurement Error in Agriculture

RCT ID

AEARCTR-0005760

First published

April 23, 2020, 6:17 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Last updated

June 15, 2020, 2:54 AM EDT

Last updated is the most recent time when changes to the trial's registration were published.

Locations

Country

India

Region

Odisha

Primary Investigator

Name

Joe Vecci

Affiliation

Gothenburg University

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Victoria Baranov

PI Affiliation

The University of Melbourne

Contact Investigator

Additional Trial Information

Status

On going

Start date

2018-04-17

End date

2020-11-02

Keywords

Agriculture

Additional Keywords

Measurement error, Agricultural production

JEL code(s)

I3, N5

Secondary IDs

Abstract

While measurement of agricultural production is critical to empirical development analyses and answering fundamental questions on the role of agriculture in household and individual welfare, there is a lack of understanding about the accuracy of farmer reported crop production and area measurement. In this project, we conduct a large, state-representative survey in Odisha, India containing self-reported and objective measures of agricultural variables to examine the extent of misreporting and measurement error in agricultural area, production, and yield. Using rich data on farmer characteristics, their psychological state, and personality traits, we explore whether misreporting and measurement error are systematically related to these characteristics. Finally, we conduct priming experiments to test if measurement error is affected by incentives to misreport.

External Link(s)

Registration Citation

Citation

Baranov, Victoria and Joe Vecci. 2020. "Understanding Measurement Error in Agriculture." AEA RCT Registry. June 15. https://doi.org/10.1257/rct.5760-1.1

Former Citation

Baranov, Victoria and Joe Vecci. 2020. "Understanding Measurement Error in Agriculture." AEA RCT Registry. June 15. https://www.socialscienceregistry.org/trials/5760/history/70439

Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Experimental Details

Interventions

Intervention(s)

The project consists of priming experiments that aim to test the extent to which measurement error in farmer reported production/area/yield is due to intentional misreporting.
Experiment 1: Financial Incentives.
The lower bound treatment group prior to asking questions related to agricultural activity were read the following preface:“We will now ask you some questions about your land holdings and agricultural production. This information will remain anonymous and it will not influence your chances of receiving benefits from any organisation including the government or IRRI.”

The upper bound treatment group were told the following preface.“We will now ask you some questions about your land holding and agricultural production. This information may influence your chances of receiving benefits from the government and organisations including IRRI in the future.” This did not involve deception as we will randomly select 25% of these households to receive a cash grant to be distributed between 2 and 6 months after the endline.

The control were not given any preface prior to eliciting agricultural activity.

Experiment 2: Lying (misreporting for any reason)
In this experiment subjects are given a prime prior to eliciting agricultural information.

In the honest prime treatment prior to eliciting agricultural information, farmers are given a list of words that are related to honesty such as truth and integrity, subjects must then list synonyms related to each of the listed words.

In the neutral prime treatment the word list is unrelated to honesty such as home and music.

A separate control contains no prime.

Intervention Start Date

2020-02-03

Intervention End Date

2020-06-02

Primary Outcomes

Primary Outcomes (end points)

Are key outcomes are the following:
1. Measurement error in production: self-reported production based of the crop cut plot minus the plot production derived from the crop cut experiment.
2. Measurement error in area: self-reported area of the crop cut plot minus the area for the crop cut plot taken from GPS measures
3. Measurement error in yield: Calculated based on self-reported production and area, yield=production/area, minus yield calculated using the above mentioned objective measures of production and area.

Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)

Secondary Outcomes (explanation)

Experimental Design

The study was conducted in Odisha, India. The sample was selected as follows. First, we randomly selected 15 out of the 30 districts in the state. We then randomly selected 300 villages, from which we undertook a full village census. From the census we randomly selected 10 farmers in each village combining for a total sample of 3000 households prior to harvest. The sample is representative of farmers in the state.

Prior to harvest (Nov-Dec, 2019), trained enumerators conducted crop cutting of the largest plot of each farmer. Crop cutting (CC) has been recognised as the gold standard for objective agricultural production measurement since the 1950s by the Food and Agriculture Organization of the United Nations (FAO). At this time, enumerators also objectively measured plot area. In particular, a polygon of the plot is drawn using a GPS device by walking along the boundary of the plot, using geospatial software a complete polygon of the plot is then created.

After harvest (February-June), enumerators return to households and elicit farmer self reported measures of production, area and yield of the largest plot.

Experimental Design Details

Randomization Method

Randomisation took place via computer and was conducted by the International Rice Research Institute

Randomization Unit

Individual

Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters

We expect a sample size of 2400 participants after subjects drop out in part due to crop failure and migration

Sample size: planned number of observations

2400

Sample size (or number of clusters) by treatment arms

Experiment 1
Lower Bound: 540
Upward Bound: 420

Experiment 2:
Honesty Prime: 540
Neutral Prime: 540

Pure Control: 650

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

We expect a sample size of 2400 participants after subjects drop out in part due to lack of harvest, crop failure or migration. Based on the sample size, we should be able to detect a minimum effect size of 0.163-0.176 SD with 80% power at 5% significance for the four main treatments (upward bias (0.163), downward bias (0.176), honesty prime (0.163), neutral prime (0.210)). We expect that neutral prime would not be different to control, but it was included as an “active” control to ensure that the simple act of reading word lists does not impact reporting. If this is confirmed in the data that reporting under the neutral prime and control is the same for all three measures (using the difference in means), then we will pool the neutral prime with the control.

Supporting Documents and Materials

IRB